In this study, I hope to explore to what extent past and present interactions between climate and human activities will generate future impacts for the purpose of reducing catchment risk through effective landscape management. Many soil conservation and
sediment control techniques are known and widely studied, including for instance reduced tillage or zero tillage, cover crops, grass buffer strips between fields or along rivers and sediment retention ponds (Verstraeten et al., 2002). The effects of most of these techniques have been analysed using experimental field plots. However, their impact at a catchment scale is not obvious. A conservation strategy should therefore integrate a variety of suitable control techniques into a catchment management plan.
To evaluate the impact of a series of possible management scenarios, a stronger spatially distributed model is needed.
Conventional flood risk studies have therefore focused on statistically based
methodologies for the determination of extreme water levels in drainage basins at a specific point of interest. Statistics is a science of description, based on mathematical principles which identify the variation in a set of observations of a process (White, 2007). However, catchments are complex dynamic systems, where responses in the sediment delivery to human actions or meteorological events may reverberate over large distances with significant time lags (Dearing and Jones, 2003). Thus, it is desirable to explore the use of spatially distributed and process-based models to simulate the behaviour of hydro-geomorphological processes in a fluvial system over long timescales. Cellular models can incorporate many processes within a simple framework that allows continuous feedback effects to generate emergent properties.
However, there are some limitations such as grid scales, vegetation changes and soil particle size distributions. Therefore, an improved cellular automaton model which is capable of modelling amalgamative slope, fluvial and hydrological processes over different time scales is needed. The overall goal of this study is to test and apply an established cellular automata landscape evolution model (CAESAR) to study the dynamic non-linear behaviour of complex systems, past and present interactions among landscape elements and environmental controls, and potential future impacts.
It is addresses by a specific set of objectives listed below.
1.3 Objectives of the research
1) To investigate the effects of climate change and human activities (e.g. land use changes, watercress cultivation etc.) on river-catchment evolution.
2) To develop and applying a dynamic model to simulate the landscape evolution of catchments over different timescales (approximately 150 years and 5000 years).
3) To compare short term (151 years) modelled water and sediment discharge with instrumental data, empirical data (e.g. sediment flux data and magnetic data obtained from sediment records).
4) To compare modelled temporal and spatial pattern of erosion and deposition to real field observations of catchment geomorphology (e.g. the elevation
difference can be calculated and displayed by ArcMap over different periods to show the erosion and deposition conditions spatially).
5) To compare the behaviour of long term (5000 years) modelled sediment discharge with laminated lake sediment record (e.g. Zolitschka, 1998).
6) To interpret the contribution of climate and land use to sediment movement through the results from both long-term and short-term simulations.
7) To explore the role of non-linearity and self-organised criticality and their behaviour in complex river catchment system (from designed simple catchment to real dynamic systems)
8) To apply the model to simulate the nature of hydro-geomorphological processes based on a series of possible scenarios built for land use and climate changes in the coming 50-100 years.
The processes required to achieve these objectives according to the time line: from the past to the future, are configured in Figure 1.1.
1.4 Structure of the thesis
Following this introduction, the heart of this thesis concentrates on testing model simulations by comparing model outputs of water and sediment discharges to environmental behaviour from observed and lake sediment records. Once the model is calibrated and validated to be robust enough to capture the dynamics of the river system, it is driven forward to explore regional hydro-geomorphic response to likely future projections of environmental changes and hence appropriate landscape
management.
Figure 1.1 Research programme structure
Chapter 2 reviews landscape behaviour to environmental changes, previous approaches used to investigate environmental changes in complex fluvial systems and the urgent requirement for models.
Chapter 3 provides a background of numerical model development and application with specification to cellular automata models and detailed introduction to the CAESAR model, which will be used in this thesis.
Chapter 4 outlines methods used in this research, including preparation of
palaeoenvironmental data for the two catchments, creations of input files for all the simulations and initial conditions setup.
Chapter 5 illustrates modelling results from Alresford catchment for a 151 year period and the construction of chronology and magnetism properties from lake sediment. Spatial and temporal comparison between simulation outputs and environmental information is applied for model validation and to provide some insights to the hydro-geomorphic response of a chalk catchment to environmental drivers.
Chapter 6 tests the feasibility of CAESAR model for long term simulations by comparing modelled sediment with lake sediment archives over 5000 years. It also highlights the human influence on the natural environment.
Chapter 7 is the key discussion chapter that demonstrates the role of non-linearity and self-organisation in complex fluvial systems. It provides evidence for the non-linear behaviour in different fluvial systems and possible factors that control them.
Chapter 8 describes the application of the CAESAR model to Alresford catchment under future climate and land use change scenarios in the next 100 years.
Chapter 9 offers a synthesis of all key findings in the research and considers future study directions.
Chapter 2
The Complexity of River Systems – from Past to Future
2.1 Landscape behaviour to environmental changes
More and more research is drawing our attention to the perspective that a river catchment is a complex cascading system (Dearing and Jones, 2003) affected by the destructive power of human impact and nature on both local and global scales. One of the direct effects of human actions and climatic events on fluvial systems is landscape evolution, with materials such as sediment shifting from one place to another (Church, 2010) through erosional processes and subsequent mobilisation and deposition by running water, wind, glacial ice etc. (de Moor and Verstraeten, 2008).
The nature of this forcing-response system may influence landscape development, ecosystem sustainability in both temporal and spatial scales and, in turn, influence the global climate through atmospheric oceanic cycles (Dearing and Jones, 2003).
The annual report of the European Environment Agency (EEA, 2012) has
highlighted a decrease in river flows in southern and eastern Europe, increases in the frequency and magnitude of both flood and drought events, earlier flowering and harvest dates for cereal crops, increased water demand for irrigation, and reduction in forest growth in central and western Europe. It is important to recognise that
geomorphological processes in response to environmental changes are based on the hydrological cycle (e.g. the impact of precipitation on runoff) and land surface cover effects. Therefore, there is a need for adequate knowledge of system sensitivity to climate and human forcings in hydrological and geomorphic processes, for the sake of the sustainable management of drainage basins.
2.1.1 Hydrological process of flooding
Widespread flooding is a natural disaster and has long been recognized as one of the most damaging, dangerous and costly hazards in the United Kingdom. For example, the historic town of Lewes on the River Ouse in East Sussex was hit by serious
floods that devastated the town centre and caused millions of pounds worth of damage in 2000 (White, 2007). It is estimated that localised flooding in England and Wales may increase by up to four-fold by 2080 (Burningham et al., 2008), primarily as a result of global warming. As a consequence, roughly five million people and two million properties will be at a risk of flooding. In this sense, flood risk is an issue of considerable concern and rational risk assessment has become one of the most important challenges that currently face river basin managers in the UK (Macklin and Rumsby, 2007).
A flood begins when main channel water levels are sufficient to exceed local bank height (Lane et al., 2007). Thus, flood risk is driven by changes of rainfall in the river channel, which may impact on both flow magnitude and river channel conveyance. In Bronstert’s (2003) view, floods can be divided into two categories according to the size of the affected area and the duration of precipitation. Extensive or plain floods are caused by rainfall lasting several days or even weeks in
connection with high antecedent soil saturation. Flooding results from extensive and long-lasting rainfall, partly related to the melting of snow and ice, occurs mostly on plains and in large areas when the dikes along the main rivers can no longer contain the flood discharge. Local or flash floods are a kind of flooding mainly caused by short, high intensity precipitation (e.g. thunderstorms) in small catchments. Flash floods occur primarily in hilly or mountainous areas due to prevailing convective rainfall mechanisms, thin soils, and high runoff velocities. The duration of this type of flood event is short, but it frequently causes severe damage as a result of the intensive rainfall and short warning time for these events. Although the risk of flooding is mainly concentrated in lowland regions, catchment headwaters, with their generally higher precipitation rates and quicker responses, are also important source areas for runoff generation (Marshall et al., 2009).
It is generally accepted that increased precipitation resulting from climate change (White, 2007) and sea level rise, coupled with pressure from increased urbanization and land use changes (Brown and Damery, 2002) as a consequence of human activities, are critical factors for flooding, which is therefore likely to become more frequent and severe in the future. Yin and Li (2001) attributed the magnitude of flooding and its long duration to the detrimental human intervention within the river
basin, which included vegetation destruction and soil erosion in the upper reaches;
decrease of the flood storage capacity due to land reclamation and siltation; and the construction of levees that caused flood levels to rise due to restricted flood
discharge capacity. Marshall et al. (2009) focused their work on the impact of modern agricultural practices, especially grassland management, on the increasing risk of flooding based on changes in soil structure and runoff processes (O'Connell et al. 2007). There is also evidence that climate change could be responsible for
increases in the magnitude of peak flows (Middelkoop et al., 2001; Milley et al., 2002) and flood frequency (Hunt 2002). On the basis of a wide array of proxy sources, such as ice cores, tree rings, corals, and lake sediments, it has been concluded that the 20th century is at least as warm as any other century of the last millennium, and may be even warmer (Bronstert, 2003). Along with global warming, increasingly heavy precipitation at continental and global scales supports the view that the global hydrological cycle is intensifying (Huntington 2006), forcing Europe into a relatively flood-rich period.
When attempting to justify flood defence works or improvements in monitoring and prediction, traditional flood management strategies have either overlooked the important social dimensions of public hazard understanding and vulnerability, or they have incorporated these factors through inappropriate quantitative measures, such as cost-benefit analysis. Instead, there is a need to adopt long-term risk
management strategies grounded in an understanding of exposure to the flood hazard, characteristics and patterns of vulnerability, and the relationships between different stakeholders in the perception of flood risk (Brown and Damery, 2002).
2.1.2 Geomorphic process of soil erosion
Soil is essential for human subsistence. The erosion of soil can occur at widely varying rates over fields, floodplains, water bodies, and even along a typical
landscape profile within a field (Foster, 1988). Driven by water, soil erosion can be categorised as sheet, rill, concentrated flow, gully and stream channel erosion (Foster, 1988), and occurs in three forms or stages (Zhang et al., 1996). The first stage is sheet erosion, which is principally caused by raindrop impact, and removes soil in a thin, almost imperceptible layer. Rill erosion is the development of numerous small,
eroded channels across a landscape due to surface runoff. These small channels are formed because of natural areal variations in the erosion resistance of the soil and small variations in elevation and slope. Flows from a large number of rills
concentrate in gullies during the final stage of soil erosion, and gullies are fairly permanent topographic features (Bennett, 1974; Zhang et al., 1996). Accordingly, understanding the controls on recent changes in soil erosion, sediment delivery, and sediment yield is of benefit to policy makers tasked with the management of rivers with high sediment loads (Walling, 1997).
Although soil erosion is a natural process, increasing population and economic development, associated with anthropogenic activity (such as intensified land use and forestry, overgrazing, deforestation, vegetation clearance and global climate change due to emission of greenhouse gases; Bronstert, 2003; Walling, 1997) may accelerate erosion and add sediment loads. Soil erosion and the delivery of eroded sediments to river channels can reduce soil productivity, generate downstream damage (Ritchie and McHenry, 1990) and pose substantial financial burdens on society. Problems related to soil erosion on arable land have numerous detrimental impacts, including the loss of topsoil and fertilisers, decreased crop yield (when plants are eroded or covered with sediment deposits) and accessibility (due to gullies) in the short-term, and decreased soil productivity in the long-term (Verstraeten et al., 2002, Ward et al., 2009). Sediment delivery also affects channel and floodplain morphology (e.g.
Asselman and Middelkoop, 1995; De Moor et al., 2008), the ecological functioning of floodplains (Richards et al., 2002), and sediment deposition rates in reservoirs and ponds (Verstraeten and Poesen 1999). Other problems associated with soil erosion include pollution of surface water with suspended sediment and other pollutants adsorbed to sediment particles (e.g. phosphates or heavy metals); silting of riverbeds, reservoirs and ponds requiring costly dredging operations; muddy flooding in local villages and substantial financial and psychological damage resulting from public infrastructure and provide property (e.g. Boardman, 2010; Verstraeten and Poesen, 1999; Verstraeten et al., 2002).
2.2 Understanding the past from lake sediment records
An understanding of past environmental changes that have occurred over previous decades and centuries is undoubtedly crucial to enable greater understanding of present and future landscape evolution and system assessment under the influence of human and natural drivers such as climatic fluctuations and shifts in land use.
Learning from the past through comprehensive palaeo-environmental archives could contribute to establishing the trajectories of forcing and response that led to current conditions (Dearing and Jones, 2003) and the possible thresholds for sensitivity or resilience to particular climate and human impacts in the future (Dearing, 2006). It is well known that rivers are efficient conveyors of water and sediment to lakes
(Williams, 2012). Information on erosion and deposition availability, transporting energy in catchment, hydrological and geomorphological processes and landscape evolution can be discerned by exploring sediments (Dearing, 1991). Foulds et al.
(2013) investigated the change in sedimentation style of an agro-industrial alluvium (Swale catchment) in northern England by using a combination of methods,
including geomorphological mapping, sedimentological and geochemical analysis, dating controls, land use history with palynological evidence and climate proxies.
They suggested the transformation of the Swale floodplain is a reflection of regional land use and climate signals of the Anthropocene. Therefore, efforts to understand the nature of river systems often start with efficient studies of sediments using a series of approaches and technologies.
2.2.1 Dating methods
After 1950, with the development of technology (e.g. modern topographical mapping and high resolution aerial photography), earth science entered an era of
unprecedented prosperity, leading to a rapid expansion of exploration and monitoring practices for earth surface environments (Church, 2010). One significant
achievement was the development of dating methods due to an urgent requirement for precise and reliable chronology. In addition, statistics were introduced into empirical science to quantify landscape variability. Radioisotope measurement pioneered in the early 1970s provided methods for establishing high resolution sediment chronology to recover data on environmental changes stored in lake
sediments (van de Post et al., 1997; Oldfield & Appleby 1984). In addition, accurate dating is of importance in interpreting changing rates and sources of allochthonous sediments related to soil erosion (Xue & Yao, 2011; Dearing, 1991).
For dating long sedimentary sequences, radiocarbon (14C), which is produced in the atmosphere by the interaction of cosmic rays (Olsson, 1986), is appropriate for dating organic remains within the time range of about 500 to 40,000 years old (Smol, 2002). The decreasing of 14C content at the decay rate, when carbon replenishment stops because of an organism such as plant or animal dies, provides an age
measurement. Owing to the limitations of radiocarbon dating for modern sediment (younger than 300 years), the short-lived radioactive isotopes of, for instance lead (210Pb) and caesium (137Cs), are employed and developed. Lead-210, with a half-life of 22.26 years, is constantly decaying in the uranium decay series. It falls onto the soil surface or directly into lakes by precipitation or dry deposition from the atmosphere and is removed by water and deposited as sediment (Appleby, 2001).
Two simple models, referred to as constant initial concentration (CIC) and constant rate of 210Pb supply (CRS) models are widely applied to calculate sediment age (Robbins, 1978; Appleby and Oldfield, 1978; Appleby, 2001). The CIC model is useful for homogeneous sediment that has a constant rate of accumulation. The CRS model is more suitable for lakes where the sediment accumulation rate varies
inversely with the concentration of 210Pb, or where the pattern of sediment focussing has been changed. Battarbee et al. (1985) reported accelerated soil erosion following forest clearance by correlating changes in sediment accumulation with afforestation history in seven lake catchments in the UK that have similar geological and climatic conditions, based on 210Pb dates. Under some circumstances of physical or biological disturbance of surficial sediment, neither of the two models is valid, and so
independent dating evidence for certain time periods is essential to validate the 210Pb results. Radioactive fallout caesium-137, one of the artificial radionuclides released after the onset of atmospheric nuclear weapons tests in 1954, can be used as a tracer to independently date sediment records. As a consequence of such nuclear
fission, 137Cs was globally distributed in the stratosphere and redeposited to the Earth’s surface as precipitation-facilitated fallout, with a peak in fallout in 1963 (Jerry et al., 1990; Davis, 1963; Longmore, 1982). The nuclear power plant accident
137
fallout was uneven and had a limited impact on global fallout patterns (Appleby, 2001; Volchok and Chieco, 1986; Jerry et al., 1990). The spatial distribution of 137Cs has been studied by McHenry and Bubenzer (1985) to explore the movement and deposition pattern of eroded materials within watersheds. Quine and Walling (1993) have demonstrated the potential of determining the topographical controls on the variations in 137Cs derived soil erosion rates for arable fields with different soil types in lowland UK. Recent studies combining 137Cs and 210Pb dating methods have provided more reliable relationships between the age and depth of lake sediment cores for estimating soil erosion rates. Appleby (2008) reviewed studies of the widespread applicability of the fallout radionuclide method and its success in dating sediment records from lakes varying in size, sedimentation rate and environmental conditions.
In addition to these commonly used radioisotopes, spheroidal carbonaceous particles preserved in sediments, which are related to fossil fuel burning, can also be used as time markers. Fossil fuels such as coal and oil droplets are burned at temperatures approaching 1750°C to produce heat and power for energy industries (Rose, 2001).
The products of incomplete combustion of fossil fuels are porous spheroids
The products of incomplete combustion of fossil fuels are porous spheroids